I’ve been moving some amplifiers around today, listening to some that have been idle for a while and doing a little side-by-side comparison. And I’ve been reminded of a very important fact about amplifier performance.
The simple truth is that a well designed amplifier will sound good and a poorly designed one won’t. There are no magical parts, no mystic capacitors, no special transformers, and no special topologies which will mitigate this simple fact. This fact seems to get lost sometimes in the search for special components or just the “right” topology.
I was reminded of this today when I swapped out the amp I was using for my office with the 6EM7 Vertical Amplifier. For those not familiar with this amp, it is a 2W per channel SET topology taking advantage of a particular dual dissimilar triode intended as a vertical deflection oscillator and driver amplifier in televisions.
On paper, the amp is nothing special. It has reasonable specification numbers and low but ample power if using moderate to high sensitivity speakers; especially in a smaller space. But what you don’t see in the numbers is the quality of the sound it produces. I designed this amp almost 10 years ago (2014) and finished assembly in early March 2015. When I plugged this amplifier into my system today I was immediately taken aback by the rich quality of the sound it produced. It has a full register with no falloffs at either end. It is fast and detailed. And it possesses that fantastic SET sound that you only know once you’ve heard it.
So I went back to the original design and looked at my development notes. I found that the operating points were very carefully chosen with attention to dynamic performance and the capabilities of each stage; driver and power. I remembered the prototyping I did to determine what seemed to work and what didn’t. And I realized that this deceptively simple design was almost a year in the making.
Usually details like this are completely hidden when just looking at a schematic. The schematic is just the output of a process which can be either long, careful, and detailed or slipshod and haphazard. And it may not be apparent from the schematic which it was. This design process was a long and careful one to make sure that I got the best performance possible out of this little tube. And it shows in the sound of the amplifier.
So this is just a reminder for builders looking for that “next” amplifier or trying to figure out “what went wrong”. The best sound comes from a good design process and good quality (but not boutique) components. Better components may make a good design sound a little better, but they cannot fix a design which was not good to begin with.
As always, questions and comments are welcome.
I too am leery of magical components save one, the output transformer. I have always heard that your amp can only be as good as the output transformer. Do you have any experience with more exotic transformers like the ones from Monolithic Magnetics, Van derveen, Tango, Luhndal, etc? Are fancier transformers only important for feedback limitations or are there other factors at play?
Output transformers don’t have nearly the impact on amplifier performance that most believe. Like anything else, they need to be well designed and manufactured and the characteristics need to be chosen for the power stage design. However, most of the marketing by the high end transformer manufacturers is pure snake oil. Once you understand bandwidth limitations, primary inductance, and overall losses the “advantage” of the pricy transformers disappears. You just need to understand the operation of the power stage design and chose the right transformer.
As for feedback, most amplifier feedback designs range from poor to terrible. When designing a feedback function, it is important to pay attention to the basic stability criteria and to make sure that the amplifier bandwidth is properly limited. Most designs I see posted on the internet do neither of these things and are only conditionally stable. There’s nothing wrong with feedback, but you really need to understand what you’re doing to do it correctly.
I remember reading George’s (aka Tubelab) experiments with the 6av5 and a super cheap ($18 at the time) Edcor transformer. He was able to dial in performance he didn’t think would be possible via cathode feedback and ultralinear operation. He gave me the impression that different transformers need to be optimized for. Clearly he was able to get great performance out of even super cheap transformers. But even he and others I’ve seen discuss this sort of thing feel that they have fewer compromises with more capable transformers. I have never built anything for myself so I try to learn by listening to people that know what they are talking about.
I have read that some older designs, Williamson I think, required a “good” transformer in order to stay stable. Maybe with SE designs that isn’t as big a worry.
Most notorious for conditional stability are quasi-complementary power amplifiers with long tailed pair input stages. Sometimes, due to internal phase shift, NFB becomes positive at some frequencies and they become better oscillators than amplifiers. And this depends on the type of load that the speaker presents to the amplifier. Fortunately quasi-complementary is a thing of the past.
My 0.02 worth. In reality, some exotic components perform worse than standard components. With most exotic stuff, you are paying for a name and bragging rights. For all practical purposes, Edcor and Hammond transformers do just fine (most folks give Edcor a slight edge over Hammond). The only ones to watch out for are cheap offshore ebay or amazon ones.
The bottom line is that far too much emphasis gets placed on output transformers. When amplifiers have stability problems, it is NOT the fault of the output transformer. The fault lies with a poor or nonexistent feedback design. Free running amplifiers (i.e. amplifiers without feedback) only have stability problems resulting from improper bandpass design and/or very poor physical layout.
I’m really happy to hear you’re still impressed with that particular amp, since I’m in the process of building one!
I enjoyed reading your thoughts. It seems that “high-end” is as much about exclusivity as performance. The prices are there to keep certain people out. While I don’t like that side of human nature, it makes me appreciate this website because you want more people to join and be a part of the hobby. Thanks!
Yep – 100% in agreement on components. I too do.not believe in “magic” components. I find it hard to stomach all the praise lavished on boutique components by the audiophile community, even things like special cryo treated grossly overpriced 3ag line fuses making vast improvements in how a system sounds. Personally, I use general purpose semiconductors and MRO grade passive components in all of my projects. As far as design goes, I use SPICE to generate a schematic and simulate operation to give me something that works. Then it is on to the breadboard, and test and tweak time.